1 |
package dk.thoerup.traininfoservice.banedk; |
package dk.thoerup.traininfoservice.banedk; |
2 |
|
|
3 |
import java.sql.Connection; |
|
4 |
import java.sql.ResultSet; |
import java.net.URL; |
5 |
import java.sql.Statement; |
import java.net.URLEncoder; |
|
import java.util.ArrayList; |
|
6 |
import java.util.Collections; |
import java.util.Collections; |
7 |
import java.util.List; |
import java.util.Map; |
8 |
|
import java.util.logging.Logger; |
9 |
|
|
10 |
import com.gargoylesoftware.htmlunit.WebClient; |
import org.jsoup.nodes.Document; |
11 |
import com.gargoylesoftware.htmlunit.html.DomNodeList; |
import org.jsoup.nodes.Element; |
12 |
import com.gargoylesoftware.htmlunit.html.HtmlElement; |
import org.jsoup.select.Elements; |
|
import com.gargoylesoftware.htmlunit.html.HtmlPage; |
|
13 |
|
|
14 |
import dk.thoerup.traininfoservice.DBConnection; |
import dk.thoerup.android.traininfo.common.DepartureBean; |
15 |
|
import dk.thoerup.android.traininfo.common.DepartureEntry; |
16 |
|
import dk.thoerup.android.traininfo.common.StationBean.StationEntry; |
17 |
|
import dk.thoerup.circuitbreaker.CircuitBreaker; |
18 |
|
import dk.thoerup.circuitbreaker.CircuitBreakerManager; |
19 |
|
import dk.thoerup.traininfoservice.StationDAO; |
20 |
|
import dk.thoerup.traininfoservice.Statistics; |
21 |
|
|
22 |
public class DepartureFetcher { |
public class DepartureFetcher { |
23 |
|
|
24 |
|
enum TrainType{ |
25 |
|
STOG, |
26 |
|
REGIONAL |
27 |
|
} |
28 |
|
|
29 |
|
enum FetchTrainType { |
30 |
|
STOG, |
31 |
|
REGIONAL, |
32 |
|
BOTH |
33 |
|
} |
34 |
|
|
35 |
|
Logger logger = Logger.getLogger(DepartureFetcher.class.getName()); |
36 |
|
|
37 |
|
Map<String, DepartureBean> cache; |
38 |
|
|
39 |
|
StationDAO stationDao = new StationDAO(); |
40 |
|
|
41 |
|
private boolean useAzureSite; |
42 |
|
private int replyTimeout; |
43 |
|
|
44 |
|
public DepartureFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) { |
45 |
|
this.replyTimeout = replyTimeout; |
46 |
|
useAzureSite = azureSite; |
47 |
|
cache = new TimeoutMap<String,DepartureBean>(cacheTimeout); |
48 |
|
} |
49 |
|
|
50 |
|
|
51 |
|
|
52 |
|
|
53 |
|
public DepartureBean cachedLookupDepartures(int stationID, boolean arrival, FetchTrainType type) throws Exception { |
54 |
|
|
55 |
|
final String key = "" + stationID + ":" + arrival + ":" + type.toString(); |
56 |
|
|
57 |
|
DepartureBean departureBean = cache.get(key); |
58 |
|
|
|
public List<DepartureBean> lookupDepartures(int stationID) throws Exception { |
|
|
List<DepartureBean> departureList = new ArrayList<DepartureBean>(); |
|
59 |
|
|
60 |
Connection conn = null; |
if (departureBean == null) { |
61 |
try |
departureBean = lookupDepartures(stationID, arrival, type); |
62 |
{ |
cache.put(key, departureBean); |
63 |
conn = DBConnection.getConnection(); |
} else { |
64 |
|
Statistics.getInstance().incrementDepartureCacheHits(); |
65 |
String SQL = "SELECT stationcode_fjrn, stationcode_stog FROM trainstations WHERE id=" + stationID; |
logger.info("Departure: Cache hit " + key); //remove before production |
|
Statement stmt = conn.createStatement(); |
|
|
ResultSet rs = stmt.executeQuery(SQL); |
|
|
|
|
|
if (rs.next()) { |
|
|
String code = rs.getString( 1 ); |
|
|
if (! rs.wasNull() ) { |
|
|
List<DepartureBean> list = lookupDepartures(code, "FJRN"); |
|
|
departureList.addAll(list); |
|
|
} |
|
|
|
|
|
code = rs.getString(2); |
|
|
if (! rs.wasNull() ) { |
|
|
List<DepartureBean> list = lookupDepartures(code, "S2"); |
|
|
departureList.addAll(list); |
|
|
} |
|
|
Collections.sort( departureList ); |
|
|
|
|
|
} |
|
|
|
|
|
} finally { |
|
|
if (conn != null && !conn.isClosed() ) { |
|
|
conn.close(); |
|
|
} |
|
66 |
} |
} |
67 |
|
return departureBean; |
68 |
|
} |
69 |
|
|
70 |
|
|
71 |
|
public DepartureBean lookupDepartures(int stationID, boolean arrival, FetchTrainType type) throws Exception { |
72 |
|
|
73 |
return departureList; |
DepartureBean departureBean = new DepartureBean(); |
74 |
|
|
75 |
|
StationEntry station = stationDao.getById(stationID); |
76 |
|
|
77 |
|
departureBean.stationName = station.getName(); |
78 |
|
|
79 |
|
if (station.getRegional() != null && (type == FetchTrainType.REGIONAL||type == FetchTrainType.BOTH) ) { |
80 |
|
DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival); |
81 |
|
departureBean.entries.addAll( tempBean.entries ); |
82 |
|
departureBean.notifications.addAll(tempBean.notifications); |
83 |
|
} |
84 |
|
|
85 |
|
if (station.getStrain() != null && (type == FetchTrainType.STOG||type == FetchTrainType.BOTH)) { |
86 |
|
DepartureBean tempBean = lookupDepartures(station.getStrain(), TrainType.STOG, arrival); |
87 |
|
departureBean.entries.addAll( tempBean.entries ); |
88 |
|
departureBean.notifications.addAll(tempBean.notifications); |
89 |
|
} |
90 |
|
|
91 |
|
if (departureBean.entries.size() == 0) { |
92 |
|
logger.info("No departures found for station " + stationID); |
93 |
|
} |
94 |
|
|
95 |
|
Collections.sort( departureBean.entries ); |
96 |
|
|
97 |
|
|
98 |
|
return departureBean; |
99 |
} |
} |
100 |
|
|
101 |
public List<DepartureBean> lookupDepartures(String stationcode, String type) throws Exception { |
public DepartureBean lookupDepartures(String stationcode, TrainType type, boolean arrival) throws Exception { |
102 |
|
if (useAzureSite == true) { |
103 |
List<DepartureBean> departureList = new ArrayList<DepartureBean>(); |
return lookupDeparturesAzureSite(stationcode, type, arrival); |
104 |
|
} else { |
105 |
|
return lookupDeparturesWwwSite(stationcode, type, arrival); |
106 |
|
} |
107 |
|
} |
108 |
|
|
109 |
|
private String getTypeStringAzure(TrainType type) { |
110 |
|
switch (type) { |
111 |
|
case STOG: |
112 |
|
return "S-Tog"; |
113 |
|
case REGIONAL: |
114 |
|
return "Fjerntog"; |
115 |
|
default: |
116 |
|
return ""; //Can not happen |
117 |
|
} |
118 |
|
} |
119 |
|
|
120 |
|
private String getTypeStringWww(TrainType type) { |
121 |
|
switch (type) { |
122 |
|
case STOG: |
123 |
|
return "S2"; |
124 |
|
case REGIONAL: |
125 |
|
return "FJRN"; |
126 |
|
default: |
127 |
|
return ""; //Can not happen |
128 |
|
} |
129 |
|
} |
130 |
|
|
131 |
|
public DepartureBean lookupDeparturesAzureSite(String stationcode, TrainType type, boolean arrival) throws Exception { |
132 |
|
|
133 |
final WebClient webClient = new WebClient(); |
DepartureBean departureBean = new DepartureBean(); |
134 |
webClient.setTimeout(1000); |
|
135 |
webClient.setJavaScriptEnabled(false); |
|
136 |
|
String typeString = getTypeStringAzure(type); |
137 |
final HtmlPage page = webClient.getPage("http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode); |
String arrivalDeparture = (arrival==false) ? "Afgang" : "Ankomst"; |
138 |
|
|
139 |
HtmlElement table = page.getElementById("afgangtabel"); |
stationcode = URLEncoder.encode(stationcode,"ISO-8859-1"); |
140 |
DomNodeList<HtmlElement> tableRows = table.getElementsByTagName("tr"); |
|
141 |
|
String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning"; |
142 |
for (HtmlElement currentRow : tableRows) { |
|
143 |
String rowClass = currentRow.getAttribute("class"); |
logger.fine("URI: " + uri); |
144 |
if (rowClass != null && rowClass.toLowerCase().contains("station") ) { |
JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), replyTimeout); |
145 |
DomNodeList<HtmlElement> fields = currentRow.getElementsByTagName("td"); |
CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); |
146 |
|
|
147 |
DepartureBean departure = new DepartureBean(); |
Document page = (Document) breaker.invoke(wrapper); |
148 |
|
|
149 |
String time = fields.get(0).asText(); |
String tableName = arrival == false ? "afgangtabel" : "ankomsttabel"; |
150 |
departure.setTime(time); |
Element table = page.getElementById(tableName); |
151 |
|
|
152 |
int updated = extractUpdated( fields.get(1) ); |
if (table != null) { |
153 |
departure.setUpdated(updated); |
Elements tableRows = table.getElementsByTag("tr"); |
154 |
|
|
155 |
String trainNumber = fields.get(2).asText(); |
//boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0); |
156 |
departure.setTrainNumber(trainNumber); |
//boolean passedTidsstreg = false; |
157 |
|
|
158 |
String destination = fields.get(3).asText(); |
for (Element currentRow : tableRows) { |
159 |
departure.setDestination(destination); |
String rowClass = currentRow.attr("class"); |
160 |
|
/* |
161 |
String origin = fields.get(4).asText(); |
if (tidsstregExists == true && passedTidsstreg == false) { |
162 |
departure.setOrigin(origin); |
if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) { |
163 |
|
passedTidsstreg = true; |
164 |
String location = fields.get(5).asText(); |
} else { |
165 |
departure.setLocation(location); |
continue; |
166 |
|
} |
167 |
String status = fields.get(6).asText(); |
}*/ |
168 |
departure.setStatus(status); |
|
169 |
|
if (rowClass != null && rowClass.toLowerCase().contains("station") ) { |
170 |
String note = fields.get(7).asText(); |
|
171 |
departure.setNote(note); |
Elements fields = currentRow.getElementsByTag("td"); |
172 |
|
|
173 |
departureList.add(departure); |
DepartureEntry departure = new DepartureEntry(); |
174 |
|
|
175 |
|
String time = fields.get(0).text(); |
176 |
|
if (time.equals("")) |
177 |
|
time = "0:00"; //Bane.dk bug work-around |
178 |
|
departure.setTime(time); |
179 |
|
|
180 |
|
int updated = extractUpdated( fields.get(1) ); |
181 |
|
departure.setUpdated(updated); |
182 |
|
|
183 |
|
String trainNumber = fields.get(2).text(); |
184 |
|
if (type == TrainType.STOG) //If it is S-train we need to extract the trainNumber |
185 |
|
trainNumber = trainNumber + " " + extractTrainNumberAzure(fields.get(2)); |
186 |
|
departure.setTrainNumber(trainNumber); |
187 |
|
|
188 |
|
String destination = fields.get(3).text(); |
189 |
|
departure.setDestination(destination); |
190 |
|
|
191 |
|
String origin = fields.get(4).text(); |
192 |
|
departure.setOrigin(origin); |
193 |
|
|
194 |
|
String location = fields.get(5).text(); |
195 |
|
departure.setLocation(location); |
196 |
|
|
197 |
|
String status = fields.get(6).text().trim(); |
198 |
|
departure.setStatus(status); |
199 |
|
|
200 |
|
String note = extractNote( fields.get(7) ); |
201 |
|
departure.setNote(note); |
202 |
|
|
203 |
|
departure.setType(typeString); |
204 |
|
|
205 |
|
departureBean.entries.add( departure ); |
206 |
|
} |
207 |
|
} |
208 |
|
} else { |
209 |
|
logger.warning("No departures found for station=" + stationcode + ", type=" + type); |
210 |
|
} |
211 |
|
|
212 |
|
Element notifDiv = page.getElementById("station_planlagte_text"); |
213 |
|
if (notifDiv != null) { |
214 |
|
|
215 |
|
Elements tables = notifDiv.getElementsByTag("table"); |
216 |
|
for (Element tab : tables) { |
217 |
|
|
218 |
|
Elements anchors = tab.getElementsByTag("a"); |
219 |
|
if (anchors.size() == 2) { |
220 |
|
departureBean.notifications.add( anchors.get(1).text() ); |
221 |
|
} |
222 |
} |
} |
223 |
|
|
224 |
|
} |
225 |
|
|
226 |
|
|
227 |
|
return departureBean; |
228 |
|
} |
229 |
|
|
230 |
|
|
231 |
|
|
232 |
|
public static String cleanText(String input) { |
233 |
|
//apparently JSoup translates characters on www.bane.dk to 0xA0 |
234 |
|
return input.replace((char) 0xA0, (char)0x20).trim(); |
235 |
|
} |
236 |
|
|
237 |
|
public DepartureBean lookupDeparturesWwwSite(String stationcode, TrainType trainType, boolean arrival) throws Exception { |
238 |
|
|
239 |
|
DepartureBean departureBean = new DepartureBean(); |
240 |
|
|
241 |
|
String type = getTypeStringWww(trainType); |
242 |
|
|
243 |
|
stationcode = URLEncoder.encode(stationcode, "ISO-8859-1"); |
244 |
|
|
245 |
|
|
246 |
|
String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode; |
247 |
|
logger.fine("URI:" + uri); |
248 |
|
|
249 |
|
|
250 |
|
JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), replyTimeout); |
251 |
|
CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); |
252 |
|
|
253 |
|
Element page = (Element) breaker.invoke(wrapper); |
254 |
|
|
255 |
|
String tableName = arrival == false ? "afgangtabel" : "ankomsttabel"; |
256 |
|
Element table = page.getElementById(tableName); |
257 |
|
|
258 |
|
|
259 |
|
|
260 |
|
if (table != null) { |
261 |
|
Elements tableRows = table.getElementsByTag("tr"); |
262 |
|
|
263 |
|
//boolean passedTidsstreg = false; |
264 |
|
//boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0); |
265 |
|
|
266 |
|
for (Element currentRow : tableRows) { |
267 |
|
String rowClass = currentRow.attr("class"); |
268 |
|
/* |
269 |
|
if (tidsstregExists == true && passedTidsstreg == false) { |
270 |
|
if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) { |
271 |
|
passedTidsstreg = true; |
272 |
|
} else { |
273 |
|
continue; |
274 |
|
} |
275 |
|
}*/ |
276 |
|
|
277 |
|
|
278 |
|
if (rowClass != null && rowClass.toLowerCase().contains("station") ) { |
279 |
|
Elements fields = currentRow.getElementsByTag("td"); |
280 |
|
|
281 |
|
DepartureEntry departure = new DepartureEntry(); |
282 |
|
|
283 |
|
|
284 |
|
|
285 |
|
String time = cleanText( fields.get(0).getAllElements().get(2).text() ); |
286 |
|
if (time.equals("")) |
287 |
|
time = "0:00"; //Bane.dk bug work-around |
288 |
|
departure.setTime(time); |
289 |
|
|
290 |
|
int updated = extractUpdated( fields.get(1) ); |
291 |
|
departure.setUpdated(updated); |
292 |
|
|
293 |
|
String trainNumber = cleanText( fields.get(2).text() ); |
294 |
|
if (type.equalsIgnoreCase("S2")) //If it is S-train we need to extract the trainNumber |
295 |
|
trainNumber = trainNumber + " " + extractTrainNumberWww(fields.get(2)); |
296 |
|
departure.setTrainNumber(trainNumber); |
297 |
|
|
298 |
|
String destination = cleanText( fields.get(3).text() ); |
299 |
|
departure.setDestination(destination); |
300 |
|
|
301 |
|
String origin = cleanText( fields.get(4).text() ); |
302 |
|
departure.setOrigin(origin); |
303 |
|
|
304 |
|
String location = cleanText( fields.get(5).text() ); |
305 |
|
departure.setLocation(location); |
306 |
|
|
307 |
|
String status = cleanText( fields.get(6).text() ); |
308 |
|
departure.setStatus(status); |
309 |
|
|
310 |
|
String note = cleanText( extractNote( fields.get(7) ) ); |
311 |
|
departure.setNote(note); |
312 |
|
|
313 |
|
departure.setType(type); |
314 |
|
|
315 |
|
departureBean.entries.add(departure); |
316 |
|
|
317 |
|
|
318 |
|
} |
319 |
|
} |
320 |
|
} else { |
321 |
|
logger.warning("No departures found for station=" + stationcode + ", type=" + type); |
322 |
} |
} |
323 |
|
|
324 |
return departureList; |
|
325 |
|
return departureBean; |
326 |
} |
} |
327 |
|
|
328 |
|
|
329 |
private int extractUpdated(HtmlElement updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif" |
private int extractUpdated(Element updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif" |
330 |
int updated = -1; |
int updated = -1; |
331 |
|
|
332 |
DomNodeList<HtmlElement> updatedImgs = updatedTd.getElementsByTagName("img"); |
Elements updatedImgs = updatedTd.getElementsByTag("img"); |
333 |
String updatedStr = updatedImgs.get(0).getAttribute("src"); |
String updatedStr = updatedImgs.get(0).attr("src"); |
334 |
|
|
335 |
if (updatedStr != null) { |
if (updatedStr != null) { |
336 |
for (int i=0; i<updatedStr.length(); i++) { |
for (int i=0; i<updatedStr.length(); i++) { |
344 |
return updated; |
return updated; |
345 |
} |
} |
346 |
|
|
347 |
|
private String extractNote(Element noteTd) { |
348 |
|
String note = noteTd.text().trim(); |
349 |
|
|
350 |
|
|
351 |
|
Elements elems = noteTd.getElementsByClass("bemtype"); |
352 |
|
if (elems.size() > 0 && note.charAt(note.length()-1) == 'i') |
353 |
|
note = note.substring(0,note.length() -1 ); |
354 |
|
|
355 |
|
return note.trim(); |
356 |
|
} |
357 |
|
|
358 |
|
private String extractTrainNumberAzure(Element trainTd) { |
359 |
|
Element anchorElement = trainTd.getElementsByTag("a").get(0); |
360 |
|
String href = anchorElement.attr("href"); |
361 |
|
|
362 |
|
int pos = href.lastIndexOf('/'); |
363 |
|
String number = href.substring(pos+1); |
364 |
|
|
365 |
|
return number; |
366 |
|
} |
367 |
|
|
368 |
|
private String extractTrainNumberWww(Element trainTd) { |
369 |
|
String number = ""; |
370 |
|
Element anchorElement = trainTd.getElementsByTag("a").get(0); |
371 |
|
String href = anchorElement.attr("href"); |
372 |
|
String argstring = href.substring( href.indexOf('?') + 1); |
373 |
|
|
374 |
|
String args[] = argstring.split("&"); |
375 |
|
for (String arg : args) { |
376 |
|
String pair[] = arg.split("="); // Key=pair[0], Value=pair[1] |
377 |
|
|
378 |
|
if (pair[0].equalsIgnoreCase("TogNr")) |
379 |
|
number = pair[1]; |
380 |
|
} |
381 |
|
|
382 |
|
|
383 |
|
return number; |
384 |
|
} |
385 |
|
|
386 |
|
|
387 |
//test |
//test |
388 |
public static void main(String args[]) throws Exception{ |
/* |
389 |
|
public static void main(String args[]) throws Exception { |
390 |
DepartureFetcher f = new DepartureFetcher(); |
DepartureFetcher f = new DepartureFetcher(); |
391 |
List<DepartureBean> deps = f.lookupDepartures("AR", "FJRN"); |
List<DepartureBean> deps = f.lookupDepartures("AR", "FJRN"); |
392 |
for(DepartureBean d : deps) { |
for(DepartureBean d : deps) { |
395 |
} |
} |
396 |
|
|
397 |
System.out.println("--------------------------"); |
System.out.println("--------------------------"); |
398 |
} |
}*/ |
399 |
} |
} |